Abstract: To solve the uncertainty and diversity of user interests during service recommendation,a service process recommendation method based on user multiple interests was proposed.The method was divided into two parts:①initial interest guidance.The N-gram model was used to learn the context of known service processes when initially creating a service process,and the context sequences were used to reduce the recommendation space for providing users with more accurate service components;②user interest extraction.Facing the diversification of user interests,probabilistic latent semantic analysis trained the user's interest-service process distribution and recommended other service processes in line with the user’s current interests.Simulation experiments showed that the proposed method could quickly and accurately recommend relevant service components and service processes to users.

Key words: user generated service, N-gram model, service component, probabilistic latent semantic analysis, service process

摘要： 为解决服务推荐过程中,用户兴趣的不确定性问题和多样性问题,提出一种基于用户多兴趣的服务流程推荐方法。该方法分为两部分:①初始兴趣引导:在初始创建服务流程时,面对用户兴趣的不确定性,利用N元模型学习已知服务流程的上下文,通过上下文顺序缩小推荐空间,为用户提供更加精准的服务链接组件;②用户兴趣抽取:在创建服务流程结束后,面对用户兴趣的多样化,概率潜在语义分析训练出用户的兴趣服务流程分布,为用户推荐出符合当下兴趣的其他服务流程。通过仿真实验表明,所提方法能够快速、准确地为用户推荐相关服务组件和业务流程链。

关键词: 用户创建服务, N元模型, 服务组件, 概率潜在语义分析, 服务流程